Instructions to use yeonseok-zeticai/QWEN_2.5_omni_3b_decoder with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use yeonseok-zeticai/QWEN_2.5_omni_3b_decoder with llama-cpp-python:

# !pip install llama-cpp-python

from llama_cpp import Llama

llm = Llama.from_pretrained(
	repo_id="yeonseok-zeticai/QWEN_2.5_omni_3b_decoder",
	filename="Qwen2.5-Omni-3B-decoder-F16.gguf",
)

llm.create_chat_completion(
	messages = [
		{
			"role": "user",
			"content": "What is the capital of France?"
		}
	]
)

Notebooks
Google Colab
Kaggle
Local Apps

llama.cpp

How to use yeonseok-zeticai/QWEN_2.5_omni_3b_decoder with llama.cpp:

Install from brew

brew install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf yeonseok-zeticai/QWEN_2.5_omni_3b_decoder:F16
# Run inference directly in the terminal:
llama-cli -hf yeonseok-zeticai/QWEN_2.5_omni_3b_decoder:F16

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama-server -hf yeonseok-zeticai/QWEN_2.5_omni_3b_decoder:F16
# Run inference directly in the terminal:
llama-cli -hf yeonseok-zeticai/QWEN_2.5_omni_3b_decoder:F16

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf yeonseok-zeticai/QWEN_2.5_omni_3b_decoder:F16
# Run inference directly in the terminal:
./llama-cli -hf yeonseok-zeticai/QWEN_2.5_omni_3b_decoder:F16

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf yeonseok-zeticai/QWEN_2.5_omni_3b_decoder:F16
# Run inference directly in the terminal:
./build/bin/llama-cli -hf yeonseok-zeticai/QWEN_2.5_omni_3b_decoder:F16

Use Docker

docker model run hf.co/yeonseok-zeticai/QWEN_2.5_omni_3b_decoder:F16

LM Studio
Jan

vLLM

How to use yeonseok-zeticai/QWEN_2.5_omni_3b_decoder with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "yeonseok-zeticai/QWEN_2.5_omni_3b_decoder"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "yeonseok-zeticai/QWEN_2.5_omni_3b_decoder",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/yeonseok-zeticai/QWEN_2.5_omni_3b_decoder:F16

Ollama
How to use yeonseok-zeticai/QWEN_2.5_omni_3b_decoder with Ollama:
```
ollama run hf.co/yeonseok-zeticai/QWEN_2.5_omni_3b_decoder:F16
```

Unsloth Studio

How to use yeonseok-zeticai/QWEN_2.5_omni_3b_decoder with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for yeonseok-zeticai/QWEN_2.5_omni_3b_decoder to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for yeonseok-zeticai/QWEN_2.5_omni_3b_decoder to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for yeonseok-zeticai/QWEN_2.5_omni_3b_decoder to start chatting

Docker Model Runner
How to use yeonseok-zeticai/QWEN_2.5_omni_3b_decoder with Docker Model Runner:
```
docker model run hf.co/yeonseok-zeticai/QWEN_2.5_omni_3b_decoder:F16
```

Lemonade

How to use yeonseok-zeticai/QWEN_2.5_omni_3b_decoder with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull yeonseok-zeticai/QWEN_2.5_omni_3b_decoder:F16

Run and chat with the model

lemonade run user.QWEN_2.5_omni_3b_decoder-F16

List all available models

lemonade list

Qwen2.5-Omni-3B Decoder (GGUF)

Text-only decoder extracted from Qwen/Qwen2.5-Omni-3B.

Architecture

Type: Qwen2VL (text decoder)
Parameters: 3.4B (decoder only, excluding vision/audio/talker/token2wav)
Hidden size: 2048
Layers: 36
Attention heads: 16 (KV heads: 2, GQA)
FFN size: 11008
Vocab: 151,936
Context: 32,768 tokens

Files

File	Size	Description
`Qwen2.5-Omni-3B-decoder-F16.gguf`	6.4 GB	Full precision (FP16)

Usage with llama.cpp

llama-cli -m Qwen2.5-Omni-3B-decoder-F16.gguf -p "Hello" -n 100 -no-cnv

Extraction

Extracted using convert_hf_to_gguf.py from llama.cpp. The converter automatically strips thinker. prefix and drops vision/audio/talker/token2wav components, keeping only the text decoder (435 tensors).

Downloads last month: 10

GGUF

Model size

3B params

Architecture

qwen2vl

Hardware compatibility

16-bit